Align Tx/Rx descriptors on 32 bytes boundary instead of PAGE_SIZE.
Also align setup descriptor on 32 bytes boundary. Tx buffer have no
alignment limitation so create dmamap without alignment
restriction[1]. Rx buffer still seems to require 4 bytes alignment
limitation but we can simply use MCLBYTES for size to map the
buffer instead of TULIP_DATA_PER_DESC as the buffer is allocated
with m_getcl(9).
de(4) supports up to TULIP_MAX_TXSEG segments for Tx buffers,
increase maximum dma segment size to TULIP_MAX_TXSEG * MCLBYTES.
While I'm here remove TULIP_DATA_PER_DESC as it is not used anymore.
This should fix de(4) breakage introduced after r176206.
Do not call BUS_DRIVER_ADDED() for detached buses (attach failed) on
driver load. This fixes crash on atapicam module load on systems,
where some ata channels (usually ata1) was probed, but failed to attach.
Before calling vflush(FORCECLOSE) mark file system as unmounted so the
following vnops will fail. This is very important, because without this change
vnode could be reclaimed at any point, even if we increased usecount. The only
way to ensure that vnode won't be reclaimed was to lock it, which would be very
hard to do in ZFS without changing a lot of code. With this change simply
increasing usecount is enough to be sure vnode won't be reclaimed from under
us. To be precise it can still be reclaimed but we won't be able to see it,
because every try to enter ZFS through VFS will result in EIO.
The only function that cannot return EIO, because it is needed for vflush() is
zfs_root(). Introduce ZFS_ENTER_NOERROR() macro that only locks
z_teardown_lock and never returns EIO.
Close race in zfs_zget(). We have to increase usecount first and then
check for VI_DOOMED flag. Before this change vnode could be reclaimed
between checking for the flag and increasing usecount.
o introduce PCIE_REGMAX and use it instead of ad-hoc constant
o where 'reg' parameter/variable is not already unsigned, cast it to
unsigned before comparison with maximum value to cut off negative
values
o use PCI_SLOTMAX in several places where 31 or 32 were explicitly used
o drop redundant check of 'bytes' in i386 pciereg_cfgread() - valid
values are already checked in the subsequent switch
rwlock implemented from libthr need to fall through the 'hard path' and
query umtx also if the shared waiters bit is set on a shared lock.
The writer starvation avoidance technique, infact, can lead to shared
waiters on a shared lock which can bring to a missed wakeup and thus
to a deadlock if the right bit is not checked (a notable case is the
writers counterpart to be handled through expired timeouts).
Fix that by checking for the shared waiters bit also when unlocking the
shared locks.
That bug was causing a reported MySQL deadlock.
Many thanks go to Nick Esborn and his employer DesertNet which provided
time and machines to identify and fix this issue.
PR: thread/135673
Reported by: Nick Esborn <nick at desert dot net>
Tested by: Nick Esborn <nick at desert dot net>
Reviewed by: jeff
- Use spin lock instead of default mutex for safety. INT/IRET instructions
save/clear/restore flags but emulated flags have no effect on the host.
I believe BIOS writers never meant to run their code in emulated
environment with interrupt enabled. :-)
- Use memcpy(3) instead of copying individual members. I believe struct
x86regs was intentionally copied from the first half of struct x86emu_regs
for this very purpose.
- Fix some style nits and consistencies.
Extract the code to find and map the MADT ACPI table during early kernel
startup and genericize it so it can be reused to map other tables as well:
- Add a routine to walk a list of ACPI subtables such as those used in the
APIC and SRAT tables in the MI acpi(4) driver.
- Move the routines for mapping and unmapping an ACPI table as well as
mapping the RSDT or XSDT and searching for a table with a given signature
out into acpica_machdep.c for both amd64 and i386.
Update 802.11s mesh support to draft 3.03. This includes a revised frame
format for peering and changes to the PERR frames.
Note that this is incompatible with the previous code.
- Split the logic to parse an SMAP entry out into a separate function on
amd64 similar to i386. This fixes a bug on amd64 where overlapping
entries would not cause the SMAP parsing to stop.
- Change the SMAP parsing code to do a sorted insertion into physmap[]
instead of an append to support systems with out-of-order SMAP entries.
PR: amd64/138220
Reported by: James R. Van Artsdalen james of jrv org
MFC after: 3 days
Don't reread the command register to see if enabling I/O or memory
decoding "took". Other OS's that I checked do not do this and it breaks
some amdpm(4) devices. Prior to 7.2 we did not honor the error returned
when this failed anyway, so this in effect restores previous behavior.
PR: kern/137668
Tested by: Aurelien Mere aurelien.mere amc-os.com
MFC after: 3 days
marius [Tue, 22 Sep 2009 11:47:21 +0000 (11:47 +0000)]
- Add missing bus_dmamap_sync(9) calls for the work DMA map. Previously
the work area was totally unsynchronized which means this driver only
had a chance of working on x86 when no bounce buffers were involved,
which isn't that likely given that support for 64-bit DMA is currently
broken throughout ata(4).
- Add necessary little-endian conversion of accesses to the work area,
making this driver work on big-endian hosts. While at it, use the
alignment-agnostic byte order encoders in order to be on the safe side.
- Clear the reserved member of the SG list entries in order to be on the
safe side. [1]
marius [Tue, 22 Sep 2009 11:38:45 +0000 (11:38 +0000)]
- According to Linux, the ALi M5451 can do 31-bit DMA instead of just
30-bit like the reset of the controllers supported by this driver.
Actually ALi M5451 can be setup up to generate 32-bit addresses by
setting the 31st bit via the accompanying ISA bridge, which allows
it to work in sparc64 machines whose IOMMU require at least 32-bit
DMA. Even though other architectures would also benefit from 32-bit
DMA, enabling this bit is limited to sparc64 as bus_dma(9) doesn't
generally guarantee that a low address of BUS_SPACE_MAXADDR_32BIT
results in a buffer in the 32-bit range.
- According to Tatsuo YOKOGAWA's ali(4), the the DMA transfer size of
ALi M5451 is fixed to 64k and in fact using the default size of 4k
- The 4DWAVE DX and NX require the recording buffer to be 8-byte
aligned so adjust the bus_dma_tag_create(9) accordingly.
- Unlike the rest of the controllers supported by this driver, the
ALi M5451 only has 32 hardware channels instead of 64 so limit the
loop in tr_intr() accordingly. [1]
Submitted by: yongari [1]
Reviewed by: yongari (superset of what is committed)
MFC after: 3 days
ed [Tue, 22 Sep 2009 11:29:11 +0000 (11:29 +0000)]
Use an unsigned integer for storing the key code.
It seems Clang breaks when checking for SPCLKEY, which is now
0x80000000. Using an unsigned integer fixes this. This is also
consistent with other pieces of kbd/syscons code, because these also use
u_int.
Build x86bios only for i386/amd64 for now. More work is required
to make these functional on other architectures, and the current
code breaks sparc64 and powerpc.
Improve mxge watchdog routine's ability to reliably reset a failed NIC:
- Mark the link as down, so if watchdog reset fails, link watching
failover software can notice it
- Don't send MXGEFW_CMD_ETHERNET_DOWN if the NIC has been reset, it is
not needed, and will fail on a freshly reset NIC.
- Ensure the transmit routines aren't attempting to PIO write to doorbells
while the NIC is being reset.
- Download the correct f/w, rather than using the EEPROM f/w after reset.
- Export a count of the number of watchdog resets via sysctl
- Zero all f/w stats at reset. This will lead to less confusing
diagnostic output when investigating NIC failures.
If CPU happens to be in usermode when a T_RESERVED trap occured,
then trapsignal is called with ksi.ksi_signo = 0. For debugging kernels,
that should end up in panic, for non-debugging kernels behaviour is
undefined.
Do panic regardeless of execution mode at the moment of trap.
To be safe, make this module initialize earlier by specifying SI_SUB_CPU
instead of SI_SUB_KLD. Note that we don't believe that there is any
consumers of x86bios that would be triggered in the early stage of boot,
so this is more or less just a safebelt.
Collapase interrupt supporting functions to a new module, and switch from
x86emu to this new module.
This changeset also brings a fix for bugs introduced with the initial
x86emu commit, which prevents the user from using some display mode or
cause instant reboots during mode switch.
Automatically depend on x86emu when vesa or dpms is being built into
kernel. With this change the user no longer need to remember building
this option.
(semiforced commit to add comment missed in last delta)
Add a maximum response length for FCP RSPNS IUs.
Clarify some of the FC option words for setting parameters
and try and disable automatic PRLI when in target mode- this
should correct some cases of N-port topologies with 23XX cards
where we put out an illegal PRLI (in target mode only we're
not supposed to put out a PRLI).
Update find(1) man page for -L/-delete interaction.
It is a bit unfortunate that the example to delete broken symlinks now uses
rm(1), but allowing this with -delete would require fixing fts(3) to not
imply FTS_NOCHDIR if FTS_LOGICAL is given (or hacks in the -delete option).
"The escape sequence '\n' shall match a <newline> embedded in
the pattern space."
It is unclear whether this also applies to a \n embedded in a
character class. Disable the existing handling of \n in a character
class following Mac OS X, GNU sed version 4.1.5 with --posix, and
SunOS 5.10 /usr/bin/sed.
Follow POSIX (IEEE Std 1003.1, 2004 Edition) in the implementation
of the y (translate) command.
"If a backslash character is immediately followed by a backslash
character in string1 or string2, the two backslash characters shall
be counted as a single literal backslash character"
The transition to Subversion allows us to rename files without
repo-copy hacks.
Remove the test-number prefix from the name of the output files,
so that new test cases can be easily added.
Re-remove the IBM0057 ID used for PS/2 mouse controllers. The asl for the
61p includes the hotkey device as IBM0068 and the mouse as IBM0057 similar
to other systems.
Old (a.out) rtld attempts to mmap zero-length region, e.g. when bss
of the linked object is zero-length. More old code assumes that mmap
of zero length returns success.
For a.out and pre-8 ELF binaries, allow the mmap of zero length.
Extract version information from git repositories. Try to find a
corresponding svn version if this is a git-svn clone'ed repo.
This prefers svn version information over git in case a working
directory has both .svn and .git directories.
Remember to unlock the peripheral prior to notifying the user. Make some
allocations M_NOWAIT so that we don't try and sleep with a nested non-sleepable
lock.
This makes the userland scsi_target begin to function again.
Add getpagesizes(3). This functions either the number of supported page
sizes or some number of the sizes themselves. It is functionally
compatible with a function by the same name under Solaris.
ed [Sat, 19 Sep 2009 17:56:26 +0000 (17:56 +0000)]
Make the keyboard layer Unicode aware.
Just take keyent_t to use an u_int to store the Unicode codepoints.
Unfortunately the keymap is now too big to be loaded using an ioctl
argument, so change the ioctl to pick a pointer.
This change breaks kbdcontrol ABI. It doesn't break X11, because X11
doesn't do anything with syscons keymaps. It just switches the device
out of K_XLATE.